llm_server/llm/__init__.py

@@ -3,13 +3,7 @@ from llm_server.llm import oobabooga, vllm
 
 
 def get_token_count(prompt: str, backend_url: str):
-    assert isinstance(backend_url, str)
-
-    if not prompt:
-        # The tokenizers have issues when the prompt is None.
-        return 0
-    assert isinstance(prompt, str)
-
+    backend_url = cluster_config.validate_backend(backend_url)
     backend_mode = cluster_config.get_backend(backend_url)['mode']
     if backend_mode == 'vllm':
         return vllm.tokenize(prompt, backend_url)

llm_server/routes/openai/chat_completions.py

@@ -150,7 +150,6 @@ def openai_chat_completions():
                             # The worker incremented it, we'll decrement it.
                             decrement_ip_count(handler.client_ip, 'processing_ips')
                             decr_active_workers(handler.selected_model, handler.backend_url)
-                            print(len(generated_text))
 
                     return Response(generate(), mimetype='text/event-stream')
                 except Exception: